Possum monitoring using raised
leg-hold traps
SCIENCE FOR CONSERVATION 164
Malcolm D. Thomas and Jennifer A. Brown
Published by
Department of Conservation
P.O. Box 10-420
Wellington, New Zealand
Science for Conservation presents the results of investigations by DOC staff, and by contracted science
providers outside the Department of Conservation. Publications in this series are internally and
externally peer reviewed.
Publication was approved by the Manager, Science & Research Unit, Science Technology and
Information Services, Department of Conservation, Wellington.
© December 2000, Department of Conservation
ISSN 1173–2946
ISBN 0–478–22010-3
Cataloguing-in-Publication data
Thomas, M. D.
Possum monitoring using raised leg-hold traps / Malcolm D.
Thomas and Jennifer A. Brown. Wellington, N.Z. : Dept. of
Conservation, 2000.
1 v. ; 30 cm. (Science for conservation, 1173-2946 ; 164).
Includes bibliographical references.
ISBN 0478220103
1. Trichosurus vulpecula—Control. 2. Animal populations—
Mathematical models. I. Brown, J. A. (Jennifer Ann)
II. Title. Series: Science for conservation (Wellington, N.Z.) ; 164.
CONTENTS
Abstract
5
1.
Introduction
6
1.1
1.2
6
7
2.
Methods
7
2.1
2.2
7
8
8
8
9
9
9
2.3
3.
Background
Objectives
Survey to determine suitable methods
Comparisons of RTC estimates and capture characteristics
2.2.1 Study sites
2.2.2 Numbers of traps and location of trap-lines
2.2.3 Method for raised sets
2.2.4 Statistical analysis
Feasibility of using a calibration index
Results
10
3.1
3.2
10
11
11
11
13
3.3
Survey to determine suitable methods
Comparisons of RTC estimates and capture characteristics
3.2.1 RTC estimates
3.2.2 Capture characteristics
Feasibility of using a calibration index
4.
Conclusions
15
5.
Recommendations
16
6.
Acknowledgements
16
7.
References
17
4
Thomas & Brown—Possum monitoring using raised leg-hold traps
Possum monitoring using raised
leg-hold traps
Malcolm D. Thomas 1 and Jennifer A. Brown 2
1
Pest Control Research Ltd, PO Box 7223, Christchurch, New Zealand.
Biomathematics Research Centre, University of Canterbury, Private Bag 4800,
Christchurch, New Zealand.
2
ABSTRACT
The Department of Conservation currently measures possum (Trichosurus
vulpecula) population abundance with leg-hold traps set on the ground.
Protected flightless birds, such as the native weka (Gallirallus australis) and
kiwi (Apteryx spp.), are at risk of being maimed or killed if caught in these
traps. Consequently the Department has specified that traps be raised above the
ground on platforms or brackets to prevent bird captures. However, there is
concern that raised traps do not catch possums as often as the traps set on the
ground so that comparisons of population indices gained from ground and
raised sets may not be valid. Field trials were conducted in three habitat types
using ground and raised sets to determine whether differences occurred in
catch rates. The trials used the survey intensity recommended for possum
monitoring specified in the National Possum Control Agencies’ trap-catch
protocol. Also, the feasibility of using a calibration index to ‘correct’ the raisedset population indices so they can be compared to ground-set indices was
examined. The results showed that there were no significant differences in
catch rates in forest. Capture trends indicated that significant differences could
occur if sample sizes were larger. Therefore comparisons of population indices
from ground and raised sets were not considered valid, and separate protocols
for their use are recommended. A standardised protocol for raised traps is
recommended. The use of a calibration index is not feasible because the error
associated with its calculation makes it meaningless. The study highlighted the
need to include error estimates when making comparisons of population
indices. Consideration needs to be given to investigating the use of alternative
monitoring devices that do not pose a risk to native birds and give more precise
estimates of population indices by providing larger sample sizes.
Keywords: possum monitoring, population indices, calibration index, residual
trap-catch, risk avoidance.
© December 2000, Department of Conservation. This paper may be cited as:
Thomas, M.D.; Brown, J.A. 2000. Possum monitoring using raised leg-hold traps. Science for
conservation 164. 17 p.
Science for conservation 164
5
1. Introduction
The Department of Conservation (DOC), Science and Research Unit, contracted
Canterbury University and Pest Control Research to compare possum
population indices calculated from captures in leg-hold traps set on the ground
(ground sets), and raised above the ground at heights of 350 mm and 700 mm
(raised sets). An initial survey was conducted to determine the current methods
used by DOC staff for raised sets. These results were used to select the most
practical method for field trials. Field trials were conducted from January to
March 2000 in three habitat types at Fiordland, Rotorua, and Galatea.
1.1
BACKGROUND
The use of standard methods to calculate possum (Trichosurus vulpecula)
population indices is an important component of effective control of possums
in New Zealand’s native forests. The methods need to provide field managers
with indices that can be compared regionally so that the limited resources
available for control can be allocated to priority areas. A standard protocol has
been developed by the National Possum Control Agencies (NPCA 2000) with
this aim and is used by DOC throughout New Zealand. The protocol outlines a
method to estimate relative possum abundance from the proportion of ground
set leg-hold traps that catch possums. When the estimates are undertaken after
possum control they are known as the residual trap-catch (RTC). Residual refers
to the population remaining after control and is commonly targeted at 5%, but
can be as low as 0.5%. The levels specified are generally considered suitable to
obtain a significant reduction in damage to the native flora and fauna.
The use of leg-hold traps puts flightless birds such as the native weka
(Gallirallus australis) and kiwi (Apteryx spp.) at risk of capture and can either
kill or permanently maim them. Consequently, DOC has specified that traps that
are set in the areas where these birds are present must be raised above the
ground by 700 mm. Traps can be set either on platforms or brackets attached
directly to tree trunks or to sloping boards set at 38o to the ground. These
guidelines were based on trap-sets used in the possum eradication programme
on Kapiti Island (Sherley 1992) and on bait station trials carried out on kiwi in
the Wellington Zoo (Robinson, A.D. unpublished report 1983). However,
further trials found that 700 mm was not high enough to prevent the capture of
weka and a height of 1000 mm or greater has been recommended where these
birds are present (Thomson, C.; Warburton, B.; Moran, L. unpublished report
1996). It is generally perceived that traps set above ground do not capture as
many possums as traps set on the ground. In an attempt to increase possum
captures trap-heights have been lowered to 350 mm in some areas, e.g.
Waikaremoana.
Concern has been expressed about the comparability of population indices
from ground and raised sets. If they are not comparable, erroneous results could
lead to areas not being allocated resources for possum control and conservation
values being compromised. There have been requests from DOC field staff for
6
Thomas & Brown—Possum monitoring using raised leg-hold traps
research trials to be undertaken to compare population indices gained from
ground and raised sets. These have been requested to determine whether any
significant differences occur between catches and whether a calibration index
can be applied to adjust the raised-set indices so that they are comparable with
ground-set indicies.
The research question for this study was to determine whether there are
meaningful differences in trap-catches between ground and raised sets.
Differences are expected because the behaviour of possums caught in traps that
are set on the ground is likely to be different from that of possums caught in
raised traps. Consequently, there are likely to be significant differences in catch
rates and, provided the sample size is large enough (i.e. many trap-lines), these
differences could be shown to be statistically significant. However, the
question of interest for DOC is whether a difference in ground- and raised-set
catch rates will have a meaningful effect on possum monitoring results when
the level of trapping effort currently undertaken is used. Therefore, this study
was conducted using sample sizes that characterise standard DOC monitoring
operations.
1.2
OBJECTIVES
• To survey DOC staff to determine methods used for raised sets and select the
most suitable methods for field trials.
• To compare RTC estimates and capture characteristics for raised sets with
standard ground sets.
• To determine the feasibility of using a calibration index to correct RTC estimates from raised sets so they can be compared with estimates from ground
sets.
2. Methods
2.1
SURVEY TO DETERMINE SUITABLE METHODS
A total of 85 DOC area offices and field centres throughout New Zealand were
sent a questionnaire that asked the following questions:
1. Do you use raised sets for possum population monitoring? (If yes please complete the following questions, if no please answer no and return the questionnaire).
2. At what height do you place traps when monitoring possums?
3. Where do you place the flour lure in relation to the trap, i.e. above or below
the trap or both above and below?
4. What device do you use to attach the trap to the tree, e.g. Scott Board, L.
Bracket, or other?
Science for conservation 164
7
5. Do you use poles or ramps when monitoring possums to improve capture
rates? If so please state what type.
Data from the questionnaire were tabulated to determine the current usage of
raised sets for possum monitoring by DOC staff. The data were also used to
decide on what was considered the most cost-effective raising device and the
best method for placement of the flour lure when using raised trap-sets in this
study.
2.2
2.2.1
COMPARISONS OF RTC ESTIMATES AND
CAPTURE CHARACTERISTICS
Study sites
Fieldwork was conducted at three study sites of approximately 500 ha that
represented habitats where DOC commonly undertakes possum monitoring.
These were: beech forest in the Eglinton and Hollyford Valleys (Fiordland, 11–
17 December 1999), mixed podocarp forest at Rotoehu Forest (21–23 January),
and forest/pasture margin at Galatea (6–9 March 2000). The sites were chosen
because they had recently undergone possum control, using a variety of
methods, and were known to contain low possum numbers, characteristic of
areas that are monitored to determine RTC levels.
2.2.2
Numbers of traps and location of trap-lines
The number of traps and trap-lines was set to be representative of current DOC
monitoring effort, i.e. 10 trap-lines for areas of 501–1000 ha. Ten trap-lines per
study site were established along compass bearings and were located so that
they were at least 200 m apart to ensure that possum captures had minimal
influence on catch rates on adjacent lines (NPCA 2000). Each trap-line
contained 21 Victor No. 1 traps located at 20 m intervals measured with a hipchain. Trap heights were either the standard ground set, as specified in the trapcatch protocol (NPCA 2000), 700 mm as specified by DOC for use where
ground birds are present, or 350 mm identified as being used in the
questionnaire. The trap-lines contained seven groups of three traps, which were
set at the three trap heights. Trap-heights within the groups were allocated
randomly using a list of random numbers to eliminate possible biases that could
occur by setting certain trap-sets at favoured sites.
The raised trap-sets at the beech and mixed podocarp sites were located on
trees, whereas the raised trap-sets at the forest/pasture site were located on
fence posts. Traps were set for 3 fine nights giving 630 trap-nights for each set
type. They were checked daily, and notes were kept on the number of possums
captured, the number of possum escapes, the number of sprung traps, and the
number of non-targets captured, as specified by the protocol. In addition,
information on sex, maturity, bone fractures (determined by palpating the leg
bones), leg caught, and whether the trap remained on the bracket when it
captured a possum, was recorded.
8
Thomas & Brown—Possum monitoring using raised leg-hold traps
2.2.3
Method for raised sets
Metal L-brackets were chosen as the trial method for raising traps above the
ground. L-brackets that contained two prongs were manufactured specifically
for the trial. The coil springs on the trap could be pushed on to the prongs,
which held the trap firmly and allowed it to be bent so that the trap could
remain horizontal regardless of the angle of the tree. This method of attachment
allowed most traps to fall to the ground when a possum was captured. All traps
had 400 mm chains that were nailed half-way between the ground and trap so
that captured possums could fall to the ground after capture.
Flour lure, as specified in the trap-catch protocol (NPCA 2000), was used for
trap-sets. Flour placement for the ground sets remained as specified, but for the
raised sets it was placed 100 mm from the ground up to and beside the trap. A
further ‘white blaze’ was located 100 mm above the trap up to 500 mm.
2.2.4
Statistical analysis
Estimates of RTC for the ground and raised sets were calculated for each habitat
type using the method described in the trap-catch protocol (NPCA 2000). The
RTC estimates were compared to determine whether there were significant
differences within each habitat type using analysis of variance (ANOVA).
Capture characteristics, i.e. percentages of males captured, percentages of
immature possums captured, percentages with broken bones, percentages of
possums captured by the rear leg, and percentages of possum escapes
(identified from possum fur in sprung traps), were compared to determine
whether significant differences occurred between the three set types. For these
analyses, data from the three habitat types were pooled.
2.3
FEASIBILITY OF USING A CALIBRATION INDEX
The feasibility of using a calibration index to ‘correct’ RTC estimates calculated
from raised-set data so they are comparable with RTC estimates calculated from
ground-set data was examined.
Data from the field trials were used to calculate two separate indices, one from
the combined data from the two forest habitats at Fiordland and Rotoehu,
where the traps were set on trees, and the other from the data from the forest/
pasture habitat at Galatea, where the traps were set on fence posts. The indices
were calculated as the ratio of the average ground-set RTC to the average raisedset RTC. Because the ratio is an estimate of the calibration index, the degree of
uncertainty about it, or statistically, the associated measure of variance of the
ratio, was calculated (Mood et al. 1974).
The feasibility of using the index was determined by calculating a ‘corrected’
RTC estimate from raised-set data and examining the width of its 95%
confidence intervals to determine whether they are able to give managers useful
data to provide meaningful comparisons.
Science for conservation 164
9
3. Results
3.1
SURVEY TO DETERMINE SUITABLE METHODS
Thirty-three replies were received from the questionnaire and 25 (76%)
indicated that they used raised sets for possum monitoring. The most common
height used for the raised sets was 700 mm (45%) but a range of other heights
was also used (Table 1). Two respondents indicated that two heights were used
in the same operation, i.e. 700 and 350 mm, and 700 and 400 mm.
TABLE 1. HEIGHTS USED FOR RAISED SETS
HEIGHT (mm)
150
USAGE
PERCENTAGE
1
4
300
1
4
350
5
17
400
1
4
400–600
1
4
500–600
700
1
4
13
45
750
1
4
1000
2
7
Knee height
2
7
Respondents used four types of raising devices (Table 2). These were:
1. Scott Boards: i.e. a piece of 8 mm 240 × 195 mm plywood wedged on to the
tree using three nails. The trap is held firmly on the board using rubber bands
made from tyre tubes.
2. L-brackets: small brackets approximately 45 × 45 mm designed to hold the
trap either by the springs or trap-base extension so that it sits at right angles to
the tree.
3. Spikes: a metal spike that is hammered into the tree and holds the trap either
by the springs or trap-base extension.
4. Ramps: a pole or board resting at an angle against the tree. The trap is attached
at the top end of the ramp.
TABLE 2. RAISING DEVICES USED.
DEVICE
Scott Board
10
USAGE
PERCENTAGE
18
66
L-bracket
7
26
Spike
1
4
Ramp
1
4
Thomas & Brown—Possum monitoring using raised leg-hold traps
Three positions were identified for the placement of the flour lure, these were:
1. Above and below the raised set.
2. Above and below but when below the flour lure is placed to one side.
3. Above only.
Sixteen respondents (46%) placed the flour lure both above and below the
raised set and 4 (16%) placed it to one side (the opposite side to the trap-chain).
The remaining 10 respondents (38%) placed the flour lure above the trap only.
3.2
3.2.1
COMPARISONS OF RTC ESTIMATES AND
CAPTURE CHARACTERISTICS
RTC estimates
There were no significant differences in catch rates from the ground or raised
sets in the two forest habitats for Fiordland (P = 0.21) and for Rotoehu (P =
0.33). However, significantly more possums were caught in ground sets at the
forest/pasture margin site at Galatea compared with traps raised to 350 mm (P =
0.03) (Fig. 1). At Galatea traps set at 350 mm had the lowest catch rate, whereas
the catch rate was lowest for the 700 mm set-traps in the two forest sites. The
low rate for the 350 mm set-traps at Galatea could have been because traps were
set on fence posts. In the forest sites the traps were on trees and there may be
behavioural differences associated with possums climbing tree trunks
compared with climbing fence posts.
Residual Trap Catch
30
25
20
Ground
350 mm
700 mm
15
10
5
0
Fiordland
Rotoehu
Galatea
Figure 1. RTC of the three trap-set heights (error bars are 95% C.L.s).
3.2.2
Capture characteristics
Percentages of males captured
There were no significant differences in the percentages of males and females
captured in ground or raised sets (P = 0.58 for the ground sets, P = 0.99 for the
350 mm sets, and P = 0.25 for the 700 mm sets) (Fig. 2).
Science for conservation 164
11
Percent
80
70
60
50
40
30
20
10
0
Ground
350 mm
700 mm
Figure 2. Percentages of males captured in traps located at different heights. (n = 85 for ground,
n = 40 for 350 mm, and n = 39 for 700 mm heights, error bars are 95% C.L.s).
Percent
Percentages of immature animals captured
Percentages of immature animals captured in the 350 mm sets did not differ
significantly from the percentages captured in the ground sets (P = 0.45).
Significantly more immature animals were captured in the 700 mm sets than on
the ground (P < 01) (Fig. 3).
90
80
70
60
50
40
30
20
10
0
Ground
350 mm
700 mm
Figure 3. Percentages of immature possums caught in sets located at the three heights. ( n = 84
for ground, n = 40 for 300 mm, and n = 40 for 700 mm heights, error bars are 95% C.L.s).
Percentages of possums with broken bones
There were no significant differences between traps set on the ground and
raised to 350 mm in the percentage of caught possums that had broken bones (P
= 0.12). Significantly more possums received
broken bones when raised sets were used at 700
mm compared with the ground sets (P < 0.01)
(Fig. 4). Nine traps that captured possums at 350
mm and 700 mm remained attached to the
brackets and six of these possums had broken
bones most likely caused by possums being unable
to fall to the ground following capture.
Possums with
broken bones (%)
50
40
30
20
10
0
Ground
350
700
Figure 4. Percentages of possums captured that had broken
bones. (n = 84 for ground, n = 40 for 350 mm, and n = 40 for
700 mm heights, error bars are 95% C.L.s).
12
These estimates of the incidence of broken bones
are likely to be conservative, as palpation of the
leg bones was not able to detect hairline fractures
that may have been present. A comparison of
broken bones with a study that identified fractures
using X-ray in addition to palpation (Warburton
Thomas & Brown—Possum monitoring using raised leg-hold traps
1992) gave a broken bone incidence that was nearly twice this study when
using ground set Victor No. 1 traps (11% cf. 6% in this study).
Percentages of traps recording possum escapes
There were no significant differences in the percentages of possum escapes
compared with traps set on the ground for traps set at 350 mm (P = 0.48) and
for traps sets set at 700 mm (P = 0.68) (Fig. 5).
60
Escapes (%)
50
40
30
20
10
0
Ground
350 mm
700 mm
Figure 5. Percentages of traps recorded with possum escapes. (n = 6 for ground, n = 8 for 350
mm, and n = 6 for 700 mm respectively, error bars are 95% C.L.s).
Back leg captures (%)
Percentages of possums captured by the back leg
There were no significant differences in the percentages of possums captured
by their back legs in traps set on the ground and raised-traps (P = 0.08 for 350
mm and P = 0.39 for 700 mm) (Fig. 6).
80
70
60
50
40
30
20
10
0
Ground
350 mm
700 mm
Figure 6. Percentages of possums captured by their back leg for the three trap-sets. (n = 84 for
ground, n = 40 for 350 mm and n = 38 for 700 mm, error bars are 95% C.L.s).
3.3
FEASIBILITY OF USING A CALIBRATION INDEX
The calculated calibration indices and their associated estimated variances for
the forest habitat and forest/pasture margin habitat are shown in Table 3.
As an example of the use of these calibration indexes, the 700 mm data from
Fiordland is ‘corrected’ using the index 1.74 and its variance of 0.78. The data
gave an ‘uncorrected’ RTC estimate of 5.71% with a variance of 29.23. The
‘corrected’ RTC estimate is calculated as follows:
Science for conservation 164
13
TABLE 3. CALIBRATION INDICES AND VARIANCE FOR RAISED SETS.
SET HEIGHT
FOREST
Ratio to ground
Variance
FOREST/PASTURE MARGIN
Ratio to ground
Variance
350 mm
1.36
0.36
2.42
3.81
700 mm
1.74
0.78
2.06
3.07
= ‘uncorrected’ RTC × index
= 5.71% × 1.74
= 9.935%
The precision, or reliability, of the corrected RTC needs to be calculated. This can be
estimated by a 95% confidence interval. Firstly, the half width of the 95% confidence
interval for the ‘uncorrected’ RTC is calculated:
‘corrected’ RTC’
CI half-width
= t0.05/2, 9 × se
= 2.26 × √(29.23/10)
= 3.86
Therefore the limits of the ‘uncorrected’ 95% confidence interval range from
1.85% to 9.58% (5.71% ±3.86%).
To calculate the 95% confidence interval of the ‘corrected’ RTC estimate the
variance of the product of the RTC × index is used. Assuming the estimated RTC
and calibration index are independent, this is estimated using the method
described by Mood et al. 1994:
var(RTC × index) = RTC2 var(index) + index2 var(RTC) + var(index) var(RTC)
= 5.712 × 0.78 + 1.742 × 29.23 + 0.78 × 29.23
= 136.19
The half width of the 95% confidence interval of the ‘corrected’ RTC is
estimated as:
= t0.05/2, 9 × se
= 2.26 × √(136.19/10)
= 8.34
The 95% confidence interval for the corrected RTC is now from 1.59% to
18.28%. This interval is more than twice as wide as the original ‘uncorrected’
RTC 95% confidence interval. The entire uncorrected interval is contained
within the corrected interval (Fig. 7).
CI half-width
20
RTC
15
10
5
Figure 7. ‘Uncorrected’ and ‘corrected’ RTC estimates from
the Fiordland 700 mm raised set data and their associated
95% confidence intervals.
14
0
"uncorrected"
Unadjusted
Thomas & Brown—Possum monitoring using raised leg-hold traps
"corrected"
Adjusted
4. Conclusions
Despite there being no statistically significant differences in capture rates
between the ground and raised sets in the forest habitat, the trends in the
capture data suggest that possum captures reduce when trap-sets are raised. It is
likely that a larger sample size would have provided data to show that
significant reductions do occur. These conclusions are consistent with a related
study (Henderson et al. 1999) that showed that possums eat significantly less
bait from bait stations that are raised above the ground compared with the
amount eaten from bait stations at ground level. Consequently direct
comparisons of raised-set RTC estimates with ground-set RTC estimates are
likely to be invalid. However, in reality, the lack of precision and the large 95%
confidence intervals associated with RTC estimates from low-density possum
populations make comparisons between areas difficult, regardless of whether
they are obtained from raised sets or not (Brown, J.A.; Thomas, M.D.
unpublished report 2000). The use of alternative sampling devices that are able
to provide larger sample sizes would improve the precision of population
indices and increase the validity of regional comparisons (Brown, J.A.; Thomas,
M.D. unpublished report 2000). Also devices could be chosen that do not
threaten the safety of native ground birds.
The increased error associated with applying a calibration index to ‘correct’ the
RTC estimates from the raised-set data makes direct comparison with groundset RTC estimates impractical. Typically, RTC estimates from low-density
populations have wide 95% confidence intervals (Brown, J.A.; Thomas, M.D.
unpublished report 2000). The 95% confidence interval will widen further
when the calibration index and associated variance are used. When the
increased width of the 95% confidence interval around the ‘corrected’ RTC
estimate includes the estimate of the ‘uncorrected’ RTC, the effect of the
adjustment is difficult to interpret.
The problem of an increased width of the 95% confidence interval when the
calibration index is applied is due to the size of the estimated variance of the
index. The variance could be reduced if a larger study were conducted (i.e.
more trap-lines). However, the cost of such a study would probably outweigh
the advantages of having directly comparable results from ground- and raisedset possum monitoring operations. A more realistic approach would be to
specify separate target levels for ground- and raised-set monitoring. More
immediate gains in monitoring consistency could be made by ensuring raisedset monitoring operations were comparable. DOC staff are using a range of
devices and methods to set and lure raised traps and there are advantages in
standardising these methods so that results can be comparable.
The capture of significantly more immature possums in the 700 mm sets
compared with ground sets also suggests that direct comparisons between
ground and raised sets may not be valid. Differences in capture rates could be
due to some behavioural effect, e.g. immature possums may be more likely to
climb trees to investigate the raised sets compared with mature possums.
Science for conservation 164
15
The larger proportion of possums with broken bones raises humaneness issues
when using raised sets. The high proportion of possums with broken bones in
the traps that remained attached to the L-bracket suggests that the failure of the
trap to be released from the raising device is a likely cause of broken bones. In
future, the device used to elevate traps will need to be tested to ensure it
releases the trap immediately after a possum is captured if bone breakages are
to remain at the same level as recorded from ground sets.
5. Recommendations
• Comparison of RTC estimates calculated from ground and raised sets should
not be made without reference to their associated 95% confidence intervals.
• The use of a calibration index to ‘correct’ RTC estimates is not recommended
because the error associated with calculating the index is likely to make the
95% confidence intervals larger. Separate target RTCs should be specified for
ground and raised sets.
• The raising device and methods for setting raised traps, including set height,
should be standardised so that raised-set operations are comparable.
• Studies should be undertaken to investigate the use of alternative monitoring
devices that provide larger sample sizes and do not threaten native birds.
• A cost-effective raising device that gives similar incidences of broken bones as
ground-set traps should be developed and tested.
6. Acknowledgements
This work was done for the Science & Research Unit, Department of
Conservation, as Investigation no. 3253.
We thank Clare Veltman and Keith Broome (Science & Research Unit, DOC) for
advice and Alan Munn (DOC Fiordland) and Dale Williams (DOC Rotorua) for
providing study sites. Also thanks to the DOC staff who took the trouble to
reply to the questionnaire and for their interest in this subject. Philip Commins
(private research contractor) provided valuable assistance with planning and
the implementation of the project in the field and Christine Bezar gave editorial
assistance.
16
Thomas & Brown—Possum monitoring using raised leg-hold traps
7. References
Henderson, R.J.; Brown, J. A; Thomas, M.D.; McAuliffe, R.J. 1999: Use of ground-level or
elevated bait stations for possum control. Proceedings of the 52nd Plant Protection
Conference: 130–135.
Mood, A.M.; Graybill, F.A.; Boes, D.C. 1974: Introduction to the theory of statistics. 3rd edn.
McGraw-Hill, New York.
NPCA 2000: Trap-catch for monitoring possum populations. National Possum Control Agencies,
Wellington.
Sherley, G.H. 1992: Eradication of brushtail possums (Trichosurus vulpecula) on Kapiti Island, New
Zealand: Techniques and methods. Science and Research Series 46. Department of
Conservation, Wellington.
Warburton, B. 1992: Victor foot-hold traps for catching Australian brushtail possums in New
Zealand: Capture efficiency and injuries. Wildlife Society Bulletin 2: 67–73.
Science for conservation 164
17